Gemini 3 Pro vs GPT-5
Wins 6 of 8 areas
Coding · Agents · Reasoning · Facts · Images and charts · Math
Wins 2 of 8 areas
Long documents · Following instructions
Gemini 3 Pro is the stronger all-rounder.GPT-5 is cheaper and better at following instructions.
Scores updated · 59 tests both models report · How we compare
Where each one wins
Tests won in each of the eight areas we test. Each piece is one test, so a longer bar means more evidence; grey means the two scored within a point of each other.
- CodingWriting and fixing software60Gemini 3 Pro6 of 7 tests · 1 tie
- ReasoningHard problems that need careful thinking50Gemini 3 Pro5 of 5 tests
- Images and chartsUnderstanding pictures, charts and video30Gemini 3 Pro3 of 4 tests · 1 tie
- FactsGetting facts right instead of making them up31Gemini 3 Pro3 of 4 tests
- AgentsCarrying out multi-step tasks on its own20Gemini 3 Pro2 of 2 tests
- MathCompetition and research-level math10Gemini 3 Pro1 of 1 test
- Following instructionsDoing exactly what it is asked02GPT-52 of 2 tests
- Long documentsFinding answers in very long texts01GPT-51 of 1 test
Math and long documents rest on a single test each.
What it costs
Prices per million tokens, roughly 750,000 words. The bars show the cost of a million tokens read plus a million written.
GPT-5 costs 20% less for the same work.
The biggest differences
The three tests each model wins by the widest margin. Scores are out of 100.
Where Gemini 3 Pro pulls ahead
- Short factual questions, answered correctlySimpleQA Verified+22.8points ahead
- Abstract visual puzzles that people can solveARC-AGI-2+21.2points ahead
- Common-sense trick questionsSimpleBench+19.7points ahead
Where GPT-5 pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+9.3points ahead
- Keeps track of context across a multi-turn chatMulti-Challenge+3.9points ahead
- Follows unfamiliar, precisely checkable instructionsIFBench+2.7points ahead
Every test, side by side
All 59 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingGemini 3 Pro
- SciCodeGemini 3 Pro by 13.256.142.9+13.2
- SWE-bench MultilingualGemini 3 Pro by 9.76555.3+9.7
- Terminal-Bench HardGemini 3 Pro by 9.141.732.6+9.1
- LMArena · WebDevGemini 3 Pro by 22 rating points14401418+22 rating
- SWE-bench ProGemini 3 Pro by 1.543.341.8+1.5
- SWE-bench VerifiedGemini 3 Pro by 1.376.274.9+1.3
- LiveCodeBench v6tie87.487tie
AgentsGemini 3 Pro
- GDPValGemini 3 Pro by 13.234.221+13.2
- BrowseCompGemini 3 Pro by 4.359.254.9+4.3
ReasoningGemini 3 Pro
- ARC-AGI-2Gemini 3 Pro by 21.231.19.9+21.2
- SimpleBenchGemini 3 Pro by 19.776.456.7+19.7
- Humanity's Last ExamGemini 3 Pro by 11.239.728.5+11.2
- GPQA DiamondGemini 3 Pro by 5.490.885.4+5.4
- CritPtGemini 3 Pro by 3.49.15.7+3.4
FactsGemini 3 Pro
- SimpleQA VerifiedGemini 3 Pro by 22.872.950.1+22.8
- AA-Omniscience · AccuracyGemini 3 Pro by 15.555.840.3+15.5
- AA-Omniscience · Non-hallucinationGPT-5 by 9.38.517.8+9.3
- Vectara HHEM hallucination ratelower is betterGemini 3 Pro by 1.513.615.1+1.5
Images and chartsGemini 3 Pro
- LMArena · VisionGemini 3 Pro by 97 rating points13051208+97 rating
- OCRBenchv2Gemini 3 Pro by 7.963.455.5+7.9
- MMMU-ProGemini 3 Pro by 680.274.2+6
- CharXiv (RQ)tie81.481.1tie
Following instructionsGPT-5
- Multi-ChallengeGPT-5 by 3.965.769.6+3.9
- IFBenchGPT-5 by 2.770.473.1+2.7
Other results33 tests, not counted
Tests outside the eight areas. They are not counted above: several are summary scores built from other tests, or the same test under another name.
- AA Agentic IndexGemini 3 Pro by 25.55226.5+25.5
- HarmBenchGPT-5 by 25.172.597.6+25.1
- AA-OmniscienceGemini 3 Pro by 2415.3-8.7+24
- Terminal-Bench 2.0Gemini 3 Pro by 1954.235.2+19
- τ²-BenchGemini 3 Pro by 15.69882.4+15.6
- AIR-Bench 2024GPT-5 by 14.573.287.7+14.5
- OJ-Bench (cpp)Gemini 3 Pro by 12.368.556.2+12.3
- HLE (with tools)Gemini 3 Pro by 10.645.835.2+10.6
- ARC-AGI-1Gemini 3 Pro by 9.37565.7+9.3
- HMMT Feb. 2025Gemini 3 Pro by 9.297.588.3+9.2
- Artificial Analysis Coding IndexGemini 3 Pro by 8.746.537.8+8.7
- LiveCodeBenchGemini 3 Pro by 8.59283.5+8.5
- frontiermath_tier_4_v1Gemini 3 Pro by 6.318.812.5+6.3
- vectara_avg_summary_lengthGPT-5 by 60.8 rating points101.9162.7+60.8 rating
- AIME 2025Gemini 3 Pro by 5.410094.6+5.4
- matharena_visual_math_overallGemini 3 Pro by 5.484.278.8+5.4
- AA IntelligenceGemini 3 Pro by 52823+5
- metr_re_benchGPT-5 by 53540+5
- τ²-Bench (Retail)Gemini 3 Pro by 4.285.381.1+4.2
- HMMT Nov. 2025Gemini 3 Pro by 4.193.389.2+4.1
- MMLU-ProGemini 3 Pro by 390.187.1+3
- VideoMMMUGemini 3 Pro by 387.684.6+3
- metr_hcastGemini 3 Pro by 2.971.368.4+2.9
- simple_safety_testsGPT-5 by 2.397.599.8+2.3
- τ²-Bench Telecom (AA run)Gemini 3 Pro by 2.387.184.8+2.3
- anthropic_red_teamGPT-5 by 29799.1+2
- bbqGemini 3 Pro by 1.698.496.8+1.6
- vectara_factual_consistencyGemini 3 Pro by 1.586.484.9+1.5
- AIME25 no toolsGemini 3 Pro by 1.49694.6+1.4
- Arena-Hard (Creative Writing)Gemini 3 Pro by 1.493.692.2+1.4
- Arena-Hard (Hard Prompt)tie72.671.9tie
- vectara_answer_ratetie99.499.9tie
- XSTesttie97.397.1tie
Questions people ask
Which is better, Gemini 3 Pro or GPT-5?
Gemini 3 Pro wins six of the eight areas we test: coding, agents, reasoning, facts, images and charts and math. GPT-5 wins long documents and following instructions, and costs 20% less.
Which is better for coding?
Gemini 3 Pro. It wins 6 of the 7 coding tests both models report; GPT-5 wins none, and 1 is a tie.
Which is cheaper?
Gemini 3 Pro costs $2.00 per million input tokens and $12.00 per million output tokens; GPT-5 costs $1.25 and $10.00. That makes GPT-5 about 20% cheaper for the same work.
How do you compare the two?
We use the 59 benchmark tests both models have published scores on. The verdict counts the 26 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 33 are listed but not counted, because several are summary scores or repeat a test. Each score is the one shown on the model's own page.